Tuesday, July 14, 2009
Documentum/EMC offers a XML database named XDb. XProc is in fact designed to work with XDb.
Perhaps XML database is a misnomer. It really is a way of storing XML documents, without (apparently) enforcing any relational integrity constraints other than those defined by the DTD (and perhaps XLink, athough so far I don't know if XLink is declarative only). Therefore XDb and XProc work hand in hand, one allowing for the storage of XML documents, the other allowing for manipulation of those documents (and perhaps, in-place updates).
The logical design is therefore done at a superior level. The 'database' concept appears to function when various stored documents are manipulated as sets - XDb supports XQuery (preferred), also XPath and XPointer.
Each XML document is stored as a DOM.Document and can be manipulated using the standard methods (createAttribute, createTextNode, etc).
I can see a possible usage in, for example, GoogleHealth - where XDb would store well-formatted templates for charts, diagnoses, allergies, vaccines, etc, which would be populated for each patient encounter and loaded into GH.
While in normal usage write contention should not be an issue, I am curious how does XDb deal with document versioning and multiple writes against the same documents - or is the R/W pipe single throttled? (later - here it is - clicking Refresh in the Db manager while an update process was underway yielded the following error:)
Interesting XML database reference information here.